https://gcc.gnu.org/bugzilla/show_bug.cgi?id=13631
--- Comment #33 from François Dumont <fdumont at gcc dot gnu.org> --- Author: fdumont Date: Wed Dec 3 19:47:00 2014 New Revision: 218329 URL: https://gcc.gnu.org/viewcvs?rev=218329&root=gcc&view=rev Log: 2014-12-03 François Dumont <fdum...@gcc.gnu.org> PR libstdc++/13631 * include/bits/codecvt.h (codecvt<char, char, mbstate_t>): friend class std::messages<char>. (codecvt<wchar_t, char, mbstate_t>): friend class std::messages<wchar_t>. * config/locale/gnu/messages_member.h (messages<char>::do_open): Specialized. (messages<char>::do_close): Likewise. (messages<wchar_t>::do_open): Likewise. (messages<wchar_t>::do_close): Likewise. * config/locale/gnu/messages_member.cc: (messages<char>::do_open): Implement. Use bind_textdomain_codeset based on codecvt<char, char, mbstate_t>._M_c_locale_codecvt code set. Use internal cache to keep opened domain name with locale information. (messages<wchar_t>::do_open): Likewise with codecvt<wchar_t, char, mbstate_t>. (messages<char>::do_close): Implement. Clean cache information. (messages<wchar_t>::do_close): Likewise. (get_glibc_msg): New. Use dgettext rather than gettext using cached domain name associated to catalog id. (messages<char>::do_get): Use latter. (messages<wchar_t>::do_get): Likewise and use also cached locale codecvt<wchar_t, char, mbstate_t> facet to convert wchar_t default value to char and the result back to wchar_t. * testsuite/22_locale/messages/13631.cc: New. * testsuite/22_locale/messages/members/char/2.cc: Use also fr_FR locale for charset conversion to get the expected accented character. Added: trunk/libstdc++-v3/testsuite/22_locale/messages/13631.cc Modified: trunk/libstdc++-v3/ChangeLog trunk/libstdc++-v3/config/locale/gnu/messages_members.cc trunk/libstdc++-v3/config/locale/gnu/messages_members.h trunk/libstdc++-v3/include/bits/codecvt.h trunk/libstdc++-v3/testsuite/22_locale/messages/members/char/2.cc
